Satellite-based estimation of total suspended solids and chlorophyll-a concentrations for the Gold Coast Broadwater, Australia.

Satellite retrieval of total suspended solids (TSS) and chlorophyll-a (chl-a) was performed for the Gold Coast Broadwater, a micro-tidal estuarine lagoon draining a highly developed urban catchment area with complex and competing land uses. Due to the different water quality properties of the rivers and creeks draining into the Broadwater, sampling sites were grouped in clusters, with cluster-specific empirical/semi-empirical prediction models developed and validated with a leave-one-out cross validation approach for robustness. For unsampled locations, a weighted-average approach, based on their proximity to sampled sites, was developed. Confidence intervals were also generated, with a bootstrapping approach and visualised through maps. Models yielded varying accuracies (R2 = 0.40-0.75). Results show that, for the most significant poor water quality event in the dataset, caused by summer rainfall events, elevated TSS concentrations originated in the northern rivers, slowly spreading southward. Conversely, high chl-a concentrations were first recorded in the southernmost regions of the Broadwater.

Vehicle trajectory prediction and generation using LSTM models and GANs.

Vehicles' trajectory prediction is a topic with growing interest in recent years, as there are applications in several domains ranging from autonomous driving to traffic congestion prediction and urban planning. Predicting trajectories starting from Floating Car Data (FCD) is a complex task that comes with different challenges, namely Vehicle to Infrastructure (V2I) interaction, Vehicle to Vehicle (V2V) interaction, multimodality, and generalizability. These challenges, especially, have not been completely explored by state-of-the-art works. In particular, multimodality and generalizability have been neglected the most, and this work attempts to fill this gap by proposing and defining new datasets, metrics, and methods to help understand and predict vehicle trajectories. We propose and compare Deep Learning models based on Long Short-Term Memory and Generative Adversarial Network architectures; in particular, our GAN-3 model can be used to generate multiple predictions in multimodal scenarios. These approaches are evaluated with our newly proposed error metrics N-ADE and N-FDE, which normalize some biases in the standard Average Displacement Error (ADE) and Final Displacement Error (FDE) metrics. Experiments have been conducted using newly collected datasets in four large Italian cities (Rome, Milan, Naples, and Turin), considering different trajectory lengths to analyze error growth over a larger number of time-steps. The results prove that, although LSTM-based models are superior in unimodal scenarios, generative models perform best in those where the effects of multimodality are higher. Space-time and geographical analysis are performed, to prove the suitability of the proposed methodology for real cases and management services.

Crime Scene Reconstruction Using a Fully Geomatic Approach.

This paper is focused on two main topics: crime scene reconstruction, based on a geomatic approach, and crime scene analysis, through GIS based procedures. According to the experience of the authors in performing forensic analysis for real cases, the aforesaid topics will be examined with the specific goal of verifying the relationship of human walk paths at a crime scene with blood patterns on the floor. In order to perform such analyses, the availability of pictures taken by first aiders is mandatory, since they provide information about the crime scene before items are moved or interfered with. Generally, those pictures are affected by large geometric distortions, thus - after a brief description of the geomatic techniques suitable for the acquisition of reference data (total station surveying, photogrammetry and laser scanning) - it will be shown the developed methodology, based on photogrammetric algorithms, aimed at calibrating, georeferencing and mosaicking the available images acquired on the scene. The crime scene analysis is based on a collection of GIS functionalities for simulating human walk movements and creating a statistically significant sample. The developed GIS software component will be described in detail, showing how the analysis of this statistical sample of simulated human walks allows to rigorously define the probability of performing a certain walk path without touching the bloodstains on the floor.